Main
James Gammerman
I’m a data scientist with a Master’s degree in Machine Learning and 3 years’ professional experience.
Within data science my main interest is in the application of classical and modern machine learning techniques to business data. I am particularly interested in the field of natural language processing. I have also published research in the field of uncertainty estimation in machine learning.
Outside of my career my main hobbies are sports, music and learning languages.
Education
MSc (part-time), Machine Learning
Royal Holloway University
London, UK
2018 - 2016
- Grade: Distinction
- Selected techniques covered: deep learning, standard classification and regression algorithms, clustering, kernel methods, Bayesian methods, real-time machine learning
- Thesis: Predictive Maintenance with Conformal and Probabilistic Prediction: A Commercial Case Study
- Won award for best thesis
MSci, Chemistry
Imperial College
London, UK
2013 - 2009
- Grade: Upper second (2:1)
- Thesis: Computational chemical physics: heat transfer in ionic liquids
- Won award for best thesis presentation
Professional Experience
Data Scientist
Centrica
London, UK
Current - 2017
- Data collection / cleaning
- Exploratory data analysis
- Modelling
- Deployment
Business Analyst
ExxonMobil
London, UK
2017 - 2013
- Provided analytical support for company’s gas production projects in Kazakhstan & North Sea, mainly cashflow modelling
- Provided ad hoc analytical support to company’s gas traders
- Various other projects e.g. analysing gas market liquidity and investigating price patterns
Academic Publications, Talks & Teaching
I have recently started publishing research in academic journals in collaboration with my alma mater Royal Holloway University.
I have also started giving talks on various topics in machine learning.
Journal article: Multi-level conformal clustering: A distribution-free technique for clustering and anomaly detection1
Neurocomputing, Volume 397, 2020, pp. 279-291
N/A
2020
- This paper was developed from my MSc thesis project.
- We introduced a novel technique which combines clustering and anomaly detection, and outlined its advantages over classical clutering techniques.
Poster: Conformal Anomaly Detection based on Association Rules2
Proceedings of Machine Learning Research, Volume 105, 2019, pp.246-7
N/A
2019
- In this commercial application we developed a new data cleaning technique.
- It combines a rule-based machine learning technique called association rule mining with the conformal prediction framework. This allowed us to automatically identify likely errors in Centrica’s SAP database which could then be manually corrected.
Talk: Machine Learning: Progress & Prospects3
Odessa University, Ukraine
Odessa, Ukraine
2018
- Guest lecture at Data Science meetup
Data Science Writing
I have recently started a website where I make blog posts about topics in data science
#TidyTuesday: Analysing cocktail recipes4
https://www.jamesgammerman.com/post/cocktail-recipes-analysis/
N/A
2020
- Blah blah blah
- Blah blah blah
#TidyTuesday: Predicting NFL stadium attendances5
https://www.jamesgammerman.com/post/predicting-stadium-attendances-with-tidymodels/
N/A
2020
- Blah blah 2
- Blah blah blah